The proposed research is conceptualized within the theoretical framework of two-state memory, where the two states are short-term memory (STM) and long-term memory (LTM). The focus is on forgetting from LTM, with the emphasis on two aspects of this process: (1) the nature of savings for information that enters and is subsequently forgotten from LTM and (2) acquisition variables that affect subsequent forgetting from LTM. The general paradigm used to investigate savings insures that all information enters LTM during original learning, followed by a pretention interval, followed by a retention test and subsequent relearning of forgotten items. Various types of information that may be saved are investigated including semantic, acoustic, graphemic, and syntactic information. The experiments explore both simple to-be-remembered information (e.g., number-noun paired associates) and more complex to-be-remembered information (e.g., sentences and paragraphs). The studies on how acquisition variables affect forgetting from LTM explore factors such as hierarchical organization, the role of rehearsal, overlearning, list length, and variable encoding. The dependent variables of interest include both error-success protocols and response latencies.